Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: For programmers who'se program is at least close to whisky level

Author: John Lowe

Date: 13:50:00 12/21/02

Go up one level in this thread


On December 21, 2002 at 16:22:45, Uri Blass wrote:

>On December 21, 2002 at 16:08:39, John Lowe wrote:
>
>>You disappoint me Uri!
>>
>>Even my 30K amateur "no C were British" program knows about isolated, doubled,
>>backward and passed pawns.
>>
>>Perhaps the Friday before the Christmas Holiday was bad timing for your
>>unveiling of your dreams.
>>
>>Most of your audience is half cut by now..............................?
>
>Yes but my program has other knowledge that is more important and seems to have
>chances against everything(it lost matches against list4.61 but got more than
>20% inspite of always changing the first move and no pondering).
>
>I guess that after I add knowledge about pawn structure and king safety it can
>become close to the top amateurs but even without it I believe that it is
>clearly better than most of the amateurs.
>
>The problem is that I want to add it in a good way that means detecting also
>weak passed pawns(I remember that hiarcs generated weak passed pawn against
>smirin and it did not understand it).
>
>Uri

No contest Uri!

I started this program in 2000 and apart from looking for checkmate at 2 ply I
never go above one ply. I'm taking the view that there's little point in it
evaluating 10,000 positions per second if it doesn't know a good position when
it sees one.

Even with no search extensions it's winning it's fair share of games (Rebel
Dacade thrashes it every time!) against a few professional programs that will
play at 1-ply.

Over-extending without support is a relevant fault to zap at the one-ply level.

Challenge it in three years!



This page took 0 seconds to execute

Last modified: Thu, 15 Apr 21 08:11:13 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.